MARK: Model convergence issue, closed pop CMR

[ratelking]

I am using MARK (version 9.x & 10.1) for estimating the population of the study species (leopard), closed population full likelihood models with pi, p, & c. There are 79 individuals captured: 33 of them only once, 11 of them twice, 7 of them three and 28 animals have more than three captures. The traps were operational for 80 days. However, due to low capture rates of the species, we pooled 5 days of occasion into 1, thus reducing the number of occasions to 20 (pooled occasions). While running the models, I found the null model (p = c, pi) estimates are not converged since the standard error is coming very small (N hat - 79, SE - 0.000009), which is unlikely/unrealistic. There is another site with 87 individuals with similar capture patterns (single captures/multiple captures), but the estimates are more realistic, and I can use more complex models such as behaviour (Mb) or heterogeneity models (Mh/Mbh). I have tried using two different versions of the software, however, the results are not much different. Can anyone please help me with this?
The question might be a bit vague at present, but I can clarify further interactions.
Regards,